Here was a great find - it turns out that Brough Turner (the CTO of NMS) is also a blogger and he recently posted a great insight on IMS.  In fact, based on discussions I have been having with various wireless operators around the world, there has been a bit of skepticism on the actual value proposition of replacing mobile infrastructure based on circuit switching with packed switched gear - the extra bandwidth, the lack of QoS (which will come, albeit in 3GPP R6), and the handset availability are all issues that will take time.  In the meantime, most services will be based on streaming of audio and/or video, location, gaming and presence.  Instead of focusing on OPEX savings and cost justification, the early adopters of IMS will revamp their networks due to strategic purposes, and not because some ROI benefits.

But one quote of Turner's points that I really enjoyed was his comment #7, namely:

SIP (with SDP, RTP, etc.) was originally a solution for peer-to-peer communications sessions, notably peer-to-peer telephony.  Subsequently, various IETF working groups have extended the SIP family of specifications to support every kind of telephony that has ever existed, specifically including central control of dumb devices.  It's difficult to see how SIP proponents can complain that 3GPP has distorted the original vision of SIP (as a peer-to-peer protocol) when the SIP community has already done the same thing.  It sounds like the real issue is "who is in charge?"

Interestingly enough, this is happening a lot with IMS.  SIP is about smart endpoints, P2P, etc.  IMS provides for a centralized call session control (the famous CSCF, also dubbed the "IMS server"), a rich centralized database (HSS), and the separation of the transport, control and application layers.  Can the two go hand-in-hand?  Yes, but in between, as we might expect, there will be a lot of "proprietary" versions of SIP - Lucent, for instance, has its SIP-Assist technology that offers a greater capacity (by 40%) on its CSCF (Lucent Session Manager or SM) via some sophisticated "parsing and messaging techniques".  SIP purists will have a field day with this - what would someone like Henry Sinnreich say?  On the other hand, the vendors will always try to build a competitive advantage, and that is to be expected.